Fuzzy Neural Network Based User Behavior Analysis

With the rapid development of the Internet, the number of network users sharplyincreases, and more and more people use the Internet now. There are many problems withthese huge amount of network user information, on the one hand, the network attacks areincreasing, and people have realized the important of analyzing user behavior to solve thenetwork security problems; on the other hand, the effective use of the mass userinformation is an important research topic for a variety of websites, particularly for thee-commerce sites. The websites analyze user behavior to provide precise marketing andservice. Therefore, the analysis of network user behavior is very important.There are variety of algorithms and models for the study of user behavior, but theyhave some shortcomings. Intrusion detection system is a technology which protects thenetwork and the system in network security domain. But there is a high false positives rateand low detection rate in the intrusion detection system. Therefore, this paper mixestogether neural networks, fuzzy systems and subtraction clustering to make theiradvantages fusion and their disadvantages disappear. A user behavior analysis systembased on fuzzy neural network is proposed in this paper. It can effectively detect attackbehavior. Firstly, the research background and the research status in the field of networksecurity is introduced, then the paper tells the research purpose and significance. After theintroduction of the concept of fuzzy theory and artificial neural networks, we state somefusion ways of fuzzy system and neural network, and evaluate their advantages anddisadvantages. Then according to the features of user behavior, we select the adaptivefuzzy neural network and use subtractive clustering to optimize the structure of the model.On this basis, a model of user behavior analysis based on fuzzy neural network is given,and design ideas and the overall framework of the model are elaborated. In addition, thepaper details the design and implementation process of the every module of theframework. Finally, a test about the system is run in simulation experiment. The result shows thatthe system can detect network attacks in high detection rate and low false positives rate.